In experiments that use horseradish peroxidase (HRP) and tetramethyl benzidine (TMB) for tracing neural connections, the activity of tissue-bound enzyme as well as the stability of the resultant reaction product are influenced by the duration of storage, the composition of the storage medium, the type of counterstaining and even the details of histological dehydration. Furthermore, the conditions for preserving HRP activity are very different from those necessary for preserving the stability of the tetramethyl benzidine (TMB) reaction product. Thus, tissue-bound HRP activity is stable at a neutral pH, while a much lower pH, around 3.3, is required for preserving the stability of the TMB reaction product. Recent evidence indicates that the stabilization bath in sodium nitroferricyanide that was previously recommended is not necessary. However, gradual dehydration of mounted sections is essential for long-term stability. Excessive counterstaining and excessive dehydration interfere with the detection of reaction product. These considerations are pertinent to experiments using free HRP as well as to those where the enzyme has been conjugated to wheat germ agglutinin.